skip to main content

Search for: All records

Creators/Authors contains: "Reed, M."

Note: When clicking on a Digital Object Identifier (DOI) number, you will be taken to an external site maintained by the publisher. Some full text articles may not yet be available without a charge during the embargo (administrative interval).
What is a DOI Number?

Some links on this page may take you to non-federal websites. Their policies may differ from this site.

  1. While machine learning approaches are rapidly being applied to hydrologic problems, physics-informed approaches are still relatively rare. Many successful deep-learning applications have focused on point estimates of streamflow trained on stream gauge observations over time. While these approaches show promise for some applications, there is a need for distributed approaches that can produce accurate two-dimensional results of model states, such as ponded water depth. Here, we demonstrate a 2D emulator of the Tilted V catchment benchmark problem with solutions provided by the integrated hydrology model ParFlow. This emulator model can use 2D Convolution Neural Network (CNN), 3D CNN, and U-Netmore »machine learning architectures and produces time-dependent spatial maps of ponded water depth from which hydrographs and other hydrologic quantities of interest may be derived. A comparison of different deep learning architectures and hyperparameters is presented with particular focus on approaches such as 3D CNN (that have a time-dependent learning component) and 2D CNN and U-Net approaches (that use only the current model state to predict the next state in time). In addition to testing model performance, we also use a simplified simulation based inference approach to evaluate the ability to calibrate the emulator to randomly selected simulations and the match between ML calibrated input parameters and underlying physics-based simulation.« less
    Free, publicly-accessible full text available December 1, 2022
  2. Hydrologists and water managers increasingly face challenges associated with extreme climatic events. At the same time, historic datasets for modeling contemporary and future hydrologic conditions are increasingly inadequate. Machine learning is one promising technological tool for navigating the challenges of understanding and managing contemporary hydrological systems. However, in addition to the technical challenges associated with effectively leveraging ML for understanding subsurface hydrological processes, practitioner skepticism and hesitancy surrounding ML presents a significant barrier to adoption of ML technologies among practitioners. In this paper, we discuss an educational application we have developed—Sandtank-ML—to be used as a training and educational tool aimedmore »at building user confidence and supporting adoption of ML technologies among water managers. We argue that supporting the adoption of ML methods and technologies for subsurface hydrological investigations and management requires not only the development of robust technologic tools and approaches, but educational strategies and tools capable of building confidence among diverse users.« less
    Free, publicly-accessible full text available December 1, 2022
  3. Integrated hydrologic models solve coupled mathematical equations that represent natural processes, including groundwater, unsaturated, and overland flow. However, these models are computationally expensive. It has been recently shown that machine leaning (ML) and deep learning (DL) in particular could be used to emulate complex physical processes in the earth system. In this study, we demonstrate how a DL model can emulate transient, three-dimensional integrated hydrologic model simulations at a fraction of the computational expense. This emulator is based on a DL model previously used for modeling video dynamics, PredRNN. The emulator is trained based on physical parameters used in themore »original model, inputs such as hydraulic conductivity and topography, and produces spatially distributed outputs (e.g., pressure head) from which quantities such as streamflow and water table depth can be calculated. Simulation results from the emulator and ParFlow agree well with average relative biases of 0.070, 0.092, and 0.032 for streamflow, water table depth, and total water storage, respectively. Moreover, the emulator is up to 42 times faster than ParFlow. Given this promising proof of concept, our results open the door to future applications of full hydrologic model emulation, particularly at larger scales.« less
    Free, publicly-accessible full text available December 1, 2022
  4. Free, publicly-accessible full text available December 1, 2022
  5. Abstract. Topography is a fundamental input to hydrologic models criticalfor generating realistic streamflow networks as well as infiltration andgroundwater flow. Although there exist several national topographic datasetsfor the United States, they may not be compatible with gridded models thatrequire hydrologically consistent digital elevation models (DEMs). Here, wepresent a national topographic dataset developed to support griddedhydrologic simulations at 1 km and 250 m spatial resolution over the contiguousUnited States. The workflow is described step by step in two parts: (a) DEMprocessing using a Priority Flood algorithm to ensure hydrologicallyconsistent drainage networks and (b) slope calculation and smoothing toimprove drainage performance. The accuracy of the derivedmore »stream network isevaluated by comparing the derived drainage area to drainage areas reportedby the national stream gage network. The slope smoothing steps are evaluatedusing the runoff simulations with an integrated hydrologic model. Our DEMproduct started from the National Water Model DEM to ensure our finaldatasets will be as consistent as possible with this existing nationalframework. Our analysis shows that the additional processing we provideimproves the consistency of simulated drainage areas and the runoffsimulations that simulate gridded overland flow (as opposed to a networkrouting scheme). The workflow uses an open-source R package, and all outputdatasets and processing scripts are available and fully documented. All ofthe output datasets and scripts for processing are published through CyVerseat 250 m and 1 km resolution. The DOI link for the dataset is (Zhang and Condon, 2020).« less
  6. ABSTRACT We report on the detection of pulsations of three pulsating subdwarf B stars observed by the Transiting Exoplanet Survey Satellite (TESS) satellite and our results of mode identification in these stars based on an asymptotic period relation. SB 459 (TIC 067584818), SB 815 (TIC 169285097), and PG 0342 + 026 (TIC 457168745) have been monitored during single sectors resulting in 27 d coverage. These data sets allowed for detecting, in each star, a few tens of frequencies that we interpreted as stellar oscillations. We found no multiplets, though we partially constrained mode geometry by means of period spacing, which recently became a key tool in analyses ofmore »pulsating subdwarf B stars. Standard routine that we have used allowed us to select candidates for trapped modes that surely bear signatures of non-uniform chemical profile inside the stars. We have also done statistical analysis using collected spectroscopic and asteroseismic data of previously known subdwarf B stars along with our three stars. Making use of high precision trigonometric parallaxes from the Gaia mission and spectral energy distributions we converted atmospheric parameters to stellar ones. Radii, masses, and luminosities are close to their canonical values for extreme horizontal branch stars. In particular, the stellar masses are close to the canonical one of 0.47 M⊙ for all three stars but uncertainties on the mass are large. The results of the analyses presented here will provide important constrains for asteroseismic modelling.« less